Working with Docker?

Docker images too big or broken? We fix it.

Most Docker setups that land on our desk were put together quickly and then left alone for years. The base image went end of life, the Dockerfile stopped reproducing, and nobody is sure how to rebuild it. We work out what the container actually needs and put it back on a footing you can maintain.

  • Since 2015Fixing technology
  • FreeEstimates
  • US teamReal engineers

docker-build.log

still failing

What it is

Docker at a glance

Docker was released in March 2013 by a company called dotCloud, later renamed Docker, Inc. It took existing Linux kernel features such as cgroups and namespaces and wrapped them in a developer-friendly command line and image format, which is how containers became something most engineers use day to day.

The pieces people usually mean by Docker today are the Docker Engine (the daemon and CLI), BuildKit and Buildx for building images, Compose for running several services together, Docker Desktop on Mac, Windows and Linux (with a commercial licence required at companies over 250 staff or 10 million dollars in revenue), and Docker Hub as the default registry.

The container image and runtime formats are now governed by the Open Container Initiative, so images built with Docker also run under Podman, containerd and CRI-O without changes.

Type
Container platform
First released
2013
License
Apache 2.0 for the Engine, commercial for Docker Desktop at scale
Maintained by
Docker, Inc. and the Open Container Initiative
Typically used for
Local dev environments, CI builds, service packaging, deployment to Kubernetes or standalone hosts

The honest version

Docker pros and cons

No technology is all upside. This is where Docker earns its place and where it tends to cause trouble.

Where Docker is strong

  • The same build runs everywhereAn image that boots on a developer laptop will boot the same way in CI and in production, which removes a whole category of 'works on my machine' arguments.
  • A mature ecosystem around itBuildKit, Buildx and Compose are stable, well documented tools, and the registry landscape (Docker Hub, GHCR, ECR) is easy to work with from almost any pipeline.
  • A huge catalogue of ready imagesOfficial images for databases, languages and web servers mean most dependencies already have a tested starting point, which cuts a lot of setup work.
  • Dev and prod look alikeRunning the same image locally and on the server gives you honest feedback much earlier than setups where each environment is hand-built.
  • Low cost to startA single Dockerfile and a compose file is enough to get going, so small teams can adopt it without a platform team behind them.
  • Reproducible pipelinesBuilds produce an artefact with a known hash, which is far easier to promote through staging than copying files around and hoping.
  • Cross-architecture buildsBuildx can produce images for both linux/amd64 and linux/arm64 from one pipeline, which matters now that Apple Silicon laptops and Graviton servers are routine.

Where Docker bites

  • Docker Hub rate limits hurt CIAnonymous pulls are capped at 100 per six hours per IP, and shared CI runners hit that quickly. Pipelines then fail with pull errors that look like outages but are really quota.
  • Image bloat is the normA COPY . /app that also pulls in .git and node_modules is enough to turn a 200 MB image into a 2 GB one, and it is the single most common thing we see.
  • Best practices are rarely followedMulti-stage builds, pinned base images and small ordered layers are well known but routinely skipped, which is why old Dockerfiles rebuild slowly and ship far more than they need.
  • Docker Desktop costs at scaleThe licence is free for small teams but paid for larger companies, which takes finance and legal sign-off some organisations did not plan for.
  • Abandoned Dockerfiles rotA base image reaches end of life, a package repository moves, a pinned version vanishes, and a file that built fine last year now refuses to build at all.
  • Security scanning is not on by defaultNothing warns you that your image ships with a decade of known CVEs. You have to add scanning in CI yourself and act on what it says.

Sound familiar?

Common Docker problems we fix

The Docker problems that bring people to us, and what you have once each one is fixed.

Image size: 2.4 GB

The image is enormous

No multi-stage build, the whole source tree copied in, build tools left in the final layer, and secrets baked into history. Pulls and deploys both pay for it.

Slimmed to a fraction of the size

pull access denied

We can't rebuild the image

The running container exists, but the Dockerfile is lost, the registry login is gone, or the base image is no longer published.

Rebuilt from what runs today

toomanyrequests: rate limit

CI keeps hitting Docker Hub limits

Every pipeline run pulls the same base images anonymously and trips the quota partway through the morning.

Pulls routed through a cache

exited (137): OOMKilled

Containers keep being killed

No memory limits set, or limits set far below what the process actually needs under load, so the kernel ends the container without warning.

Limits that match real usage

platform mismatch arm64 vs amd64

It runs locally but not on the server

An image built on an Apple Silicon laptop quietly ships as arm64, then fails to start on an amd64 production host.

Multi-arch builds in the pipeline

last successful build: 412 days ago

The previous developer vanished

A production container nobody owns, no CI pipeline, no documentation on how it was built or where it was deployed from.

Ownership, build and deploy restored

COPY failed: file not found

The build broke and nobody knows why

A path that used to exist, a base image that changed, or a BuildKit setting that differs between laptop and CI.

Builds green and repeatable

What we do

Docker help, from rescue to upkeep

Call us for one broken build, or hand us the whole container estate.

Rescue
Builds that won't complete, containers that won't start, or images nobody can reproduce. We get them running and then put them on a maintainable footing.
Slim images
Multi-stage builds, pinned bases and layer ordering, with before-and-after sizes and pull times so the change is measurable.
Registries and rate limits
Moving off anonymous Docker Hub pulls onto a pull-through cache or your own registry, so CI stops failing on quota.
Takeover
Containers inherited from a developer who has moved on. We trace what is running, rebuild the source of truth, and hand you back the keys.
Compose and CI
Clean compose files for local development and reliable build pipelines that produce signed, scanned images.
Managed Tech
Ongoing patching, base image refreshes, scanning and a team to call when a build fails at 2am.

How it works

From first message to fixed

  1. Look Free estimate

    Tell us what's going wrong. We look at it, then tell you plainly what's wrong and what it will cost.

  2. Fix

    We fix the cause, not the symptom. If something is down, getting it back up comes first.

  3. Harden

    Backups, monitoring, updates and access, so the same problem can't quietly come back.

  4. Hand back or stay on

    You get a plain-language write-up. Or we stay on for hosting, upkeep and new work.

Often found together

Running Docker alongside something else?

We work on the whole stack, so the problem can be wherever it actually is.

Questions

Docker questions

Something else? Call+1 737 727 7327

Can you rebuild Docker images we no longer have the source for?

Often yes. We inspect the running container, pull its layers apart, recover what we can of the original build, and produce a clean Dockerfile that reproduces what you have today. Then you own the source again and can patch it.

Why is our Docker image 2GB when it should be 200MB?

Almost always a single-stage build that copies the whole repository in, including .git, node_modules, build tools and test fixtures, on top of a full operating system base. A multi-stage build with a slim runtime base routinely takes that same application under 200MB.

Can you take over Docker infrastructure from a previous developer?

Yes, and it is a lot of what we do. We find what is actually running, recover or rebuild the Dockerfiles, move the registry and CI into accounts you control, and document it so the next person can carry on without starting over.

How do we stop hitting Docker Hub rate limits on CI?

Put a pull-through cache or your own registry (GHCR, ECR, GAR, Harbor) in front of Docker Hub so repeat pulls do not count against the anonymous quota. Authenticating pulls and pinning digests helps too, and it makes builds reproducible as a side effect.

Should we move from Docker to Podman?

Only if there is a real reason, such as rootless containers on servers or the Docker Desktop licence at a large company. Podman uses the same OCI images Docker does, so migration is usually straightforward, but it is rarely urgent on its own.

How much does Docker help cost?

Estimates are free. Tell us what is broken or what you have inherited and you get a price before any work begins.

Free estimate

Docker giving you trouble?

Tell us what you're running and what's going wrong. A person reads it and comes back with an estimate.

  • No charge to find out what it will take
  • If it's urgent, say so

Replies come from a person. By sending, you authorize us to email you back. No marketing.